A profile is a written, versioned search definition — area, activity codes, company type, status and window — sized to the practice's weekly review capacity.
Historic volume tells you how many companies you would have had to review over a past window. It does not tell you how many clients you will win, or which of those companies need an accountant. Building a useful profile is mostly a capacity exercise: define the proposition, translate it into Companies House filters, measure what those filters would have returned, and adjust until the count fits what the practice can actually review each week. This guide works through that sequence with a hypothetical example, then covers how to adjust a profile once it is running.
Build a profile step by step
Each step below produces one part of the profile. For what each register field means, see the Companies House data guide.
| Step | Purpose | Owner | Input | Action | Output | Completion check | Common failure |
|---|---|---|---|---|---|---|---|
| 1. Proposition | Define who the practice wants as clients before touching any filter. | You | A description of the target client type. | Write one sentence naming the client type and what is deliberately excluded. | A written proposition statement. | The sentence names a type of client and at least one exclusion. | Describing an “ideal client” by imagined qualities (budget, ambition) rather than something a filter can test. |
| 2. Geography | Bound the search to a service area. | You | The practice's realistic service area. | Choose a registered-office address text match, a postcode area, or a local-authority description. | One geography definition, recorded as text. | The definition specifies exactly what string or boundary will be entered into a search. | Assuming an address text match behaves like a postcode-area or radius search — see the caveat in finding newly incorporated companies. |
| 3. Activity codes | Match the proposition to Companies House activity codes. | You | The written proposition. | Identify candidate SIC codes; see SIC codes explained for how to select them. | A list of candidate codes. | Each candidate code's official description matches the proposition. | Picking one broad code to save searches, rather than the code that fits — the free search takes one code per search. |
| 4. Type and status | Restrict to entities the practice can service and companies still on the register. | You | Legal forms served and desired statuses. | Select the matching company-type and status filters. | A type and status filter. | The filters match a form and status the practice can actually take on. | Leaving status unticked, which pulls in dissolved companies. |
| 5. Window | Choose how far back to look, matched to review cadence. | You | A review cadence, e.g. weekly. | Pick an incorporation-date window long enough for a stable count (commonly 30–31 days for a first test). | A defined window with start and end dates. | The window is written down, not recalled as “the last month”. | Comparing counts from windows of different lengths and reading the difference as a trend. |
| 6. Measure volume | Find out how many companies the profile would have surfaced. | You | The geography, codes, type, status and window above. | Run the search(es), as in the Companies House walkthrough, and record each count. | One or more dated historical counts. | Every count is stored with its date, window and exact filters. | Treating the count as a forecast rather than a backward-looking figure. |
| 7. Compare with capacity | Decide whether the volume fits what the practice can review. | You | The historical volume and weekly review capacity. | Convert the volume to a weekly figure and compare it with capacity. | A decision: proceed, narrow, or widen. | The comparison is written down alongside the count it is based on. | Comparing a monthly count with a weekly capacity without converting either figure. |
| 8. Record the card | Keep the definition reproducible over time. | You | Every decision from steps 1–7. | Fill in the profile card. | A versioned, dated record. | A colleague could rerun the exact same search from the card alone. | Changing the definition without a new version, so no one can tell which count belongs to which filters. |
Three versions of one hypothetical profile
Take the same hypothetical four-person Bristol practice used in the Companies House walkthrough, which wants owner-managed trades and small construction contractors and can review about 15 companies a week. One proposition can produce a specialist, balanced or broad profile depending on how many activity codes it includes. All figures below are historical 31-day counts observed on 17 September 2026, for the window 18 August to 17 September 2026, with company type Private limited company and status Active — not a forecast.
| Version | Geography | Activity codes | Observed 31-day volume | Method note |
|---|---|---|---|---|
| Specialist | Address text “Bristol”. | 43220 (plumbing, heat and air-conditioning installation) only. | 8 | One search. |
| Balanced | Address text “Bristol”. | 43210 (electrical installation), 43220, 43390 (other building completion and finishing). | 20 (5 + 8 + 7) | Three separate searches, one per code; a company filing two of these codes could appear in more than one search, so this total can double-count. |
| Broad | Address text “Bristol”. | All construction-related codes. | Not measured — bounded above by 546, the unfiltered count for the same address and window. | Measure your own broad set one code at a time; do not assume the unfiltered figure is the profile's real volume. |
The specialist and balanced counts are small enough that a single unusual week could move them noticeably; treat any count under about ten as unstable until it has been observed over more than one window.
Adjusting a profile to capacity
Convert weekly review capacity to the same period as the measured volume before comparing. A practice with a weekly capacity of 15 records can review roughly 60–65 a month (15 × four to just over four weeks). Once the volume and the capacity are in the same units, four rules keep adjustment disciplined:
- change one dimension per test — geography, one activity code, or the window, never several at once;
- run each version for at least four weeks before judging it, since a single week's count is not reliable on its own;
- version the profile card every time a filter changes, so a count can always be traced back to the exact definition that produced it;
- treat every count as historical, not as a promise of future volume or client demand.
| Signal | Likely cause | Adjustment |
|---|---|---|
| Volume clearly exceeds capacity | The profile is broader than the practice can review. | Narrow one dimension: drop to a single more specific code, tighten the geography text, or add a company-type restriction. |
| Volume is under a third of capacity | The profile is narrower than the practice needs to fill its review time. | Widen one dimension: add one more activity code, broaden the geography text, or extend the window slightly. |
| Counts swing sharply week to week | The base count is too small to be stable. | Extend the observation window before deciding anything; do not adjust off a single week. |
| Matches look unrelated to the proposition on review | The code or geography combination is mismatched, not a capacity problem. | Reconsider the activity codes against the proposition rather than narrowing by volume alone. |
A profile card to copy
Record every profile in the same format so a past version can always be reproduced and compared with the next one.
| Proposition | One sentence: target client type and exclusions. |
|---|---|
| Geography | Address text, postcode area or local-authority description, written exactly as entered. |
| Activity codes | Every SIC code included, one row per code if searched separately. |
| Company type / status | The type(s) and status(es) selected. |
| Window | Start and end date of the measured period. |
| Observed volume | The count, with the date it was measured. |
| Weekly capacity | The practice's review capacity used for comparison. |
| Decision | Proceed, narrow or widen, and why. |
| Version / date | A version number and the date this card was recorded. |
Measuring your own band with the free search
- Write your proposition and pick candidate activity codes, using SIC codes explained if the choice is unclear.
- Follow the walkthrough in finding newly incorporated companies on Companies House to run each code against your geography, type and status.
- Record each count with its date and exact filters, the same day you run it.
- Convert your weekly review capacity to the same period as the count and compare.
- Fill in the profile card above and set a date to review it again after at least four weeks.
What a profile is not
A profile is a screening filter, not a client-quality signal. Keep these boundaries in view:
- it does not identify “ideal clients” — it identifies companies matching a written definition, which is a different thing;
- a higher or lower historic volume says nothing about lead quality or how many of those companies will become clients;
- historic volume is not future volume or demand — the register changes daily, and a past count is not a guarantee;
- small counts are unstable; a profile returning single digits needs a longer observation window before any adjustment decision is trusted;
- company-level rows from a search are working data for the practice, not something to publish or share outside the review process.
How Launch Register can help
The profile card above is usable with nothing but a spreadsheet and the free search. Once it exists, Launch Register can turn it into a saved profile, give a historical 30-day volume estimate for it, and show a sample of up to 10 matching companies to review. Sizing the profile to capacity and deciding whether any match is relevant remain the practice's decisions. See how Launch Register works.
Test this profile against a free sample
See a historical volume estimate and up to 10 matching companies for the profile you have just defined.
Personalised sample of up to 10 companies · no card required · £29/month to continue